Merge pull request #7 from bglid/feature/min-disk-persistence

Feature/min disk persistence

Author: bglid

.gitignore

@@ -34,3 +34,6 @@ setup.log
 
 # Local OPAM switch
 _opam/
+
+# db file
+*.db

bin/main.ml

@@ -1,3 +1,13 @@
 let () =
   Printf.printf "*********SQCaml*********\n\r";
-  SQCaml.Repl.start ()
+  let db_dir =
+    if Array.length Sys.argv >= 2 then
+      Sys.argv.(1)
+    else
+      "sqcaml.db"
+  in
+  let db = SQCaml.Db_session.open_db db_dir in
+  try SQCaml.Repl.start db
+  with exn ->
+    SQCaml.Db_session.close_db db;
+    raise exn

lib/ast/ast.ml

@@ -12,13 +12,6 @@ type binop_t =
   | Comp
 [@@deriving show]
 
-(* type column_types = *)
-(*   | Int_col *)
-(*   | Float_col *)
-(*   | String_col *)
-(*   | Bool_col *)
-(* [@@deriving show] *)
-
 (** Our Abstract Syntax Tree for interpreter, hooray! *)
 type expr =
   | Int of int
@@ -48,10 +41,9 @@ type statement =
 type meta_command =
   | Help
   | Exit
+  | Tree
   | Unk_mcmd of string
-[@@deriving show]
 
 type top_level =
   | Statement of statement
   | Meta_command of meta_command
-[@@deriving show]

lib/ast/insert.ml

@@ -1,29 +1,49 @@
 (* abstraction to make a module for preparing and executing insert statement*)
 
 type t = {
-  table : Table.t;
+  (* tree : Btree.t; *)
   fields : string list;
   values : Constant.t list;
 }
 [@@deriving show]
 
 (* HACKY for now, new table each time *)
-let make (tablename : string) (fields : string list) (values : Constant.t list)
-    : t =
-  { table = Table.create_table ~id:777 ~table_name:tablename; fields; values }
+(* let make (tablename : string) (fields : string list) (values : Constant.t list) *)
+let make (fields : string list) (values : Constant.t list) : t =
+  (* { table = Table.create_table ~id:777 ~table_name:tablename; fields; values } *)
+  { fields; values }
 
-let execute_insert (preped_insert : t) : string =
-  (* let new_table = Table.add_row ~table:temp_table ~row:new_row in *)
-  let new_fields =
-    List.fold_left
-      (fun acc l ->
-        match l with
-        | Constant.ConstStr s -> acc ^ " " ^ s
-        | Constant.ConstInt d -> acc ^ " " ^ Int.to_string (Int32.to_int d))
-      "" preped_insert.values
+let execute_insert (db : Db_session.t) (preped_insert : t) : string =
+  (* serialize the insert to a page*)
+  let new_row =
+    match preped_insert.values with
+    | [ id; stop; rail ] ->
+        Table.make_row
+          ~id:(Int32.to_int (Constant.to_int id))
+          ~stop_name:(Constant.to_str stop) ~rail_line:(Constant.to_str rail)
+    | _ -> failwith "Wrong amount of args passed"
   in
-  (* new_fields *)
-  "Inserted" ^ new_fields ^ " into " ^ preped_insert.table.table_name
+  let cursor = Cursor.tree_end db.index in
 
-(* Printf.printf "Inserted %d %s %s into Table %d" id stop_name rail_name *)
-(*   preped_insert.table.table_name *)
+  (* check to guard capacity*)
+  let leaf = Btree.get_node db.index cursor.page_num in
+  if leaf.cur_size >= leaf.capacity then
+    failwith "HELP! please implement splitting plz."
+  else
+    (* serialize the page and write to a block! *)
+    let page =
+      Table.serialize_to_page new_row
+        ~block_size:(File_manager.get_blocksize db.file_manager)
+    in
+    let row_block =
+      Storage_manager.append ~storage_m:db.storage_manager ~page
+    in
+    (* block num where we want to write to disk!!*)
+    let row_block_num = Page.Block.block_num row_block in
+    let key = Keys.Integer (Int32.of_int new_row.id) in
+    (* getting and writing our tree node to disk FINALLY*)
+    Cursor.leaf_node_insert cursor key row_block_num;
+
+    (* printing the res *)
+    let res = "Inserted " ^ new_row.stop_name ^ " " ^ new_row.rail_line ^ " " in
+    res

lib/ast/select.ml

@@ -10,12 +10,12 @@ type t = {
 let make (fields : string list) (tablename : string) : t =
   { table = Table.create_table ~id:777 ~table_name:tablename; fields }
 
-let execute_select (preped_select : t) : string =
+let execute_select (_ : Db_session.t) (preped_select : t) : string =
   let selected_fields =
     List.fold_left
       (fun acc l ->
         match l with
-        | "id" -> acc ^ " 1"
+        (* | "id" -> acc ^ " 1" *)
         | "stop_name" -> acc ^ "englewood"
         | "rail_line" -> acc ^ " G"
         | _ -> "not a valid field in the table")

lib/ast/table.ml

@@ -15,41 +15,86 @@ type t = {
 }
 [@@deriving show]
 
+(* hard coding rows for serialization *)
+let stop_name_size = 64
+let rail_line_size = 64
+let id_offset = 0
+let stop_offset = id_offset + 4
+let rail_offset = stop_name_size + stop_offset
+let row_size = 4 + stop_name_size + rail_line_size
 let make_row ~id ~stop_name ~rail_line : row_t = { id; stop_name; rail_line }
 
 let create_table ~(id : int) ~(table_name : string) : t =
-  { table_id = id; table_name; table_fields = []; rows = [] }
+  let rows = [] in
+  { table_id = id; table_name; table_fields = []; rows }
 
 (* let open_table ~(table_name : string) : t = *)
 (*   match table_name with *)
 (*     |  *)
 
-let already_in_table ~(table : t) ~(row : row_t) : bool =
-  (* List.mem row table.rows *)
-  let rec search (table_rows : row_t list) (id : int) : bool =
-    match table_rows with
-    | [] -> false
-    | h :: tail ->
-        if h.id = id then
-          true
-        else
-          search tail id
-  in
-  search table.rows row.id
+(* let already_in_table ~(table : t) ~(row : row_t) : bool = *)
+(*   (* List.mem row table.rows *) *)
+(*   let rec search (table_rows : row_t list) (id : int) : bool = *)
+(*     match table_rows with *)
+(*     | [] -> false *)
+(*     | h :: tail -> *)
+(*         if h.id = id then *)
+(*           true *)
+(*         else *)
+(*           search tail id *)
+(*   in *)
+(*   search table.rows row.id *)
 
 (* currently this is a bit functional in it's nature
    always returning a new table rather than adjusting the existing one.
    May want to consider changing rows to be mutable, or need to consider disk 
    writes here - TBD *)
-let add_row ~(table : t) ~(row : row_t) : t =
-  if already_in_table ~table ~row then
-    table
+(* let add_row ~(table : t) ~(row : row_t) : t = *)
+(*   if already_in_table ~table ~row then *)
+(*     table *)
+(*   else *)
+(*     (* let new_row_list = List.fold_left (fun acc l -> acc :: l) table.rows [row] in *) *)
+(*     let new_row_list = table.rows @ [ row ] in *)
+(*     { *)
+(*       table_id = table.table_id; *)
+(*       table_name = table.table_name; *)
+(*       table_fields = table.table_fields; *)
+(*       rows = new_row_list; *)
+(*       num_rows = List.length new_row_list; *)
+(*     } *)
+
+let pad_fixed_size (size : int) (s : string) : string =
+  if String.length s > size then
+    failwith
+      (Printf.sprintf "String too long.\nGreater than input size of %d" size)
   else
-    (* let new_row_list = List.fold_left (fun acc l -> acc :: l) table.rows [row] in *)
-    let new_row_list = table.rows @ [ row ] in
-    {
-      table_id = table.table_id;
-      table_name = table.table_name;
-      table_fields = table.table_fields;
-      rows = new_row_list;
-    }
+    s ^ String.make (size - String.length s) '\000'
+
+let trim_null (s : string) : string =
+  match String.index_opt s '\000' with
+  | None -> s
+  | Some thing -> String.sub s 0 thing
+
+let serialize_to_page (row : row_t) ~(block_size : int) : Page.Page.t =
+  if row_size > block_size then
+    failwith "Row doesn't fit into page";
+
+  let page = Page.Page.make ~block_size in
+  (* serialize id, stop name, then rail name *)
+  Page.Page.set_int32 page id_offset (Int32.of_int row.id);
+  Page.Page.set_string_raw page stop_offset
+    (pad_fixed_size stop_name_size row.stop_name);
+  Page.Page.set_string_raw page rail_offset
+    (pad_fixed_size rail_line_size row.rail_line);
+  page
+
+let deserialize_from_page (page : Page.Page.t) : row_t =
+  (* get the id, stop name, and rail name from the Int32 offsets*)
+  let id = Int32.to_int (Page.Page.get_int32 page id_offset) in
+  let stop_name =
+    trim_null (Page.Page.get_string_raw page stop_offset stop_name_size)
+  in
+  let rail_line =
+    trim_null (Page.Page.get_string_raw page rail_offset rail_line_size)
+  in
+  { id; stop_name; rail_line }

lib/btree/btree.ml

@@ -120,3 +120,60 @@ let get_node (btree : t) (p : int) : Nodes.t =
     Storage_manager.get_block ~storage_m:btree.storage_m ~block_num:p
   in
   deserialize page btree.key block_size
+
+[@@@warning "-32"]
+
+let empty_node (btree : t) : Nodes.t =
+  let block_size = File_manager.get_blocksize btree.storage_m.file_manager in
+  let key_type = btree.key in
+  let capacity = get_num_keys block_size key_type in
+  {
+    Nodes.node_t = Leaf;
+    Nodes.parent = 0;
+    Nodes.cur_size = 0;
+    Nodes.keys = Array.init capacity (fun _ -> Keys.empty_key key_type);
+    Nodes.pointers = Array.init (capacity + 1) (fun _ -> unused_pointer_serial);
+    Nodes.capacity;
+    Nodes.key_type;
+  }
+
+(* Make an empty btree initialized to disk *)
+let create (storage_m : Storage_manager.t) (key_type : Keys.t) : t =
+  let block_size = File_manager.get_blocksize storage_m.file_manager in
+  let meta = Storage_manager.get_head_page ~storage_manager:storage_m in
+  Page.Page.set_int32 meta 4 (Int32.of_int 1);
+  Storage_manager.update_block_num ~storage_m ~block_num:0 ~page:meta;
+  let capacity = get_num_keys block_size key_type in
+  let root_node =
+    {
+      Nodes.node_t = Leaf;
+      Nodes.parent = 0;
+      Nodes.cur_size = 0;
+      Nodes.keys = Array.init capacity (fun _ -> Keys.empty_key key_type);
+      Nodes.pointers =
+        Array.init (capacity + 1) (fun _ -> unused_pointer_serial);
+      Nodes.capacity;
+      Nodes.key_type;
+    }
+  in
+  let root_page = serialize root_node block_size in
+  let _ = Storage_manager.append ~storage_m ~page:root_page in
+  { storage_m; key = key_type; root = root_node; root_num = 1 }
+
+let open_btree (storage_m : Storage_manager.t) (key_type : Keys.t) : t =
+  let block_size = File_manager.get_blocksize storage_m.file_manager in
+  let meta = Storage_manager.get_head_page ~storage_manager:storage_m in
+  let root_num = Int32.to_int (Page.Page.get_int32 meta 4) in
+  if root_num = 0 then
+    (* create btree instead *)
+    create storage_m key_type
+  else
+    let root_page = Storage_manager.get_block ~storage_m ~block_num:root_num in
+    let root_node = deserialize root_page key_type block_size in
+    { storage_m; key = key_type; root = root_node; root_num }
+
+[@@@warning "-32"]
+
+let print_tree (tree : t) : string =
+  let root_node = get_node tree tree.root_num in
+  Nodes.print_leaf_node root_node

lib/btree/btree.mli

@@ -1,4 +1,9 @@
-type t
+type t = {
+  storage_m : Storage_manager.t; (* file used to store to disk *)
+  key : Keys.t;
+  mutable root : Nodes.t;
+  mutable root_num : int;
+}
 (** Struct for the B+ Tree *)
 
 val serialize : Nodes.t -> int -> Page.Page.t
@@ -22,3 +27,9 @@ val write_node_append : t -> Nodes.t -> int
 val get_node : t -> int -> Nodes.t
 (** Get a block from the btree [t] and deserializeit using the pointer [p] into
     a btree node *)
+
+val open_btree : Storage_manager.t -> Keys.t -> t
+(** Opens btree given [storage manager] and [key type]. If it doesn't exist,
+    creates a new btree *)
+
+val print_tree : t -> string

lib/btree/nodes.ml

@@ -36,4 +36,16 @@ let int32_to_node_t (i32 : Int32.t) : node_type =
   else if i32 = internal_serial then
     Internal
   else
-    failwith "WRONG i32!"
+    failwith (Printf.sprintf "WRONG i32: %ld" i32)
+
+let print_leaf_node (n : t) : string =
+  if n.node_t <> Leaf then
+    "Not a leaf node"
+  else
+    let lines =
+      List.init n.cur_size (fun i ->
+          match n.keys.(i) with
+          | Keys.Integer n -> Printf.sprintf "- %d\n" (Int32.to_int n)
+          | Keys.Varchar v -> Printf.sprintf "- %s\n" v)
+    in
+    Printf.sprintf "leaf (size %d)\n%s" n.cur_size (String.concat "" lines)

lib/btree/nodes.mli

@@ -19,3 +19,4 @@ type t = {
 
 val serialize_node : node_type -> Int32.t
 val int32_to_node_t : Int32.t -> node_type
+val print_leaf_node : t -> string